AI Red Team Prompt Pack v1.0 — 100 LLM Security Testing Prompts

Stop guessing. Start breaking.

This is the most comprehensive, field-tested collection of 100 adversarial prompts for stress-testing LLM security — built by AI red teamers, for AI red teamers.

WHAT'S INSIDE:

100 battle-ready prompts across 18 attack categories:

- Prompt Injection (Direct & Indirect)

- Jailbreak Attacks (DAN, persona hijacking, roleplay exploits)

- System Prompt Extraction

- Multi-Turn Manipulation

- Context Window Poisoning

- Token Smuggling & Encoding Bypasses

- Goal Hijacking & Instruction Override

- Data Exfiltration via LLM

- Memory & Embedding Poisoning

- Model Denial of Service

- Adversarial Suffix Attacks

- Constitutional AI Bypasses

- Multi-Modal Injection

- Agent Task Hijacking

- Supply Chain Prompt Attacks

- Alignment Probing

- Red Team Scenario Chains

Each prompt includes:

- Category & Severity Level (Low / Medium / High / Critical)

- Attack Goal

- Expected Vulnerable Response

- Detection Indicator

WHO THIS IS FOR:

- AI Security Researchers

- Red Teamers & Penetration Testers

- LLM Application Developers

- Bug Bounty Hunters targeting AI systems

- Anyone building or auditing LLM-powered products

FORMAT: Instantly downloadable PDF — 24 pages, clean layout, ready to use in any testing workflow.

Find vulnerabilities before attackers do.